Recreational use of illicit drugs (i.e. use not associated with a diagnosed drug problem) may cause psychological and social harm. A recent systematic review found that evidence for this was equivocal. Extensive evidence was only available in relation to cannabis use. This was relatively consistently associated with lower educational attainment and greater use of other drugs. However whether this association was causal was not clear. Cannabis use was less consistently associated with mental illness and antisocial behaviour. Causal relations between cannabis use and psychosocial harm could plausibly be mediated through either neurophysiological effects of cannabis or through social mechanisms related to use of an illegal substance. These different mechanisms might have different implications for harm-reduction policy. Alternatively associations may arise through non-causal pathways such as reverse causation, bias and confounding. In this latter situation, even effective reduction of cannabis use would be unlikely to be an effective harm-reduction policy in relation to psychosocial outcomes. Research strategies that could clarify these questions are discussed, as are the implications of these considerations for harm-reduction policy.